• Great video Roy!
    Do you know any open source .net projects which have really good unit and integration tests?

  • My opinion on using var in tests is that it should only be used if the type is 100% clear (such as var myType = new MyType() ) or if the type is non-important to the operation it is being used in (such as foreach var v in Type).

    If you are testing an object of a certain type that is instantiated in a method call you should declare it as that type for this reason: If a type can be cast to many different types and its instantiation comes from a method call, how can you be sure of the type you are then testing? I think this is also a case where you don't want the refactor friendly behavior of var. The type of the actual object under test should not change silently via rafactoring IMHO. If the method call is refactored to return a different, but compatible type, I think the test should break.

  • What is the point of these video's Roy? If its public ridicule your spot on. Otherwise you need to work on displaying your professionalism. Did you need to sink to a WTF level in the video?

    If your goal is to help people learn/improve their tests then highlight an issue and show them how to fix it.

  • Great review, a bit on the long side though. Amazing what you can find in a test suite. Makes me think harder at cleaning up my tests.

    @Aaron Fisher: I aggree, I think it would be very useful if you showed more how to fix the problems that you highlight. Not everyone has read your book ;)

  • I agree with some of the other comments that it would have been more instructive on how to improve some of the tests. However, given that this was recorded very early in the morning and likely on impulse; I can fault the content too much :)

  • Thanks again for a great review. Each one helps me to reaffirm the stuff I learned from your book (which I bought based on reviews on Amazon rather than by looking at the illegally distro-ed copy of your book online.).

    I'm reading it in tandem with 'working effectively with legacy code - Michael Feathers' and I find they complement each other very well.

    More specifically about this Test Review I would also like to see you refactor some (definitely not all - as the number of tests you cover is important too) of the tests as well.

  • Cool stuff Roy, great screencast. We are looking into it and may refactor things a bit - time permitting.

Comments have been disabled for this content.